Peak-to-Average Power Ratio Analysis for OFDM-Based Mixed-Numerology Transmissions

Xiaoran Liu, Lei Zhang, Jun Xiong, Xiaoying Zhang, Li Zhou, Jibo Wei

TL;DR

As 5G moves away from "one-size-fits-all" signaling toward flexible "mixed-numerology" transmissions, traditional models for predicting signal power spikes (PAPR) have broken down. This paper introduces a new, generic mathematical framework based on level-crossing theory that accurately predicts these power peaks, providing a vital tool for engineers designing efficient 5G power amplifiers and interference-reduction schemes.

What Problem Does This Paper Address?

Orthogonal Frequency Division Multiplexing (OFDM) is the backbone of modern wireless communication, but it has a notorious "peakiness" problem. When many subcarriers align, they create a high Peak-to-Average Power Ratio (PAPR), which can distort signals or even damage power amplifiers (PAs) if not managed.

In 4G/LTE, every subcarrier used the same spacing. However, 5G New Radio (NR) introduces mixed-numerology, where different subbands use different subcarrier spacings (SCS) and cyclic prefixes to support diverse needs—like low latency for self-driving cars and high density for IoT. Existing PAPR models assume a continuous, symmetric frequency spectrum, which simply isn't true for these complex 5G "composite" signals. This gap makes it difficult for engineers to determine the necessary "back-off" for power amplifiers, leading to either wasted energy or signal distortion.

Methods

The authors move away from simple empirical approximations and instead leverage level-crossing theory. This approach treats the signal as a continuous-time random process and calculates the expected number of times the signal envelope crosses a specific power threshold.

Key aspects of their method include:

  • Generic Analytical Expression: They derived a new formula for the Complementary Cumulative Distribution Function (CCDF) of PAPR that accounts for subband-specific parameters: numerology, bandwidth, and power levels.
  • Broad Applicability: Unlike previous models, this one works for conventional single-numerology, mixed-numerology, and even Non-Continuous OFDM (NC-OFDM).
  • Power Allocation Optimization: They formulated power allocation as an optimization problem to mathematically determine which power distributions across subbands result in the highest or lowest PAPR.

Datasets

The paper relies on rigorous numerical simulations rather than external datasets to validate the theory:

  • Simulation Setup: $10^6$ randomly generated symbols using 16-QAM modulation.
  • Oversampling: Signals were 8-times oversampled to accurately approximate continuous-time behavior (standard PAPR analysis often fails if only Nyquist-rate samples are used).
  • Benchmarks: The results were compared against established models by van Nee (empirical) and Ochiai/Imai (theoretical), as well as existing bounds for NC-OFDM.

Key Insights and Ideas

  • Bandwidth is the Driver: A major "aha moment" is that PAPR is largely determined by the total bandwidth of the subbands, regardless of the specific numerology (subcarrier spacing) chosen within those bands.
  • The "Gap Filler" for NC-OFDM: The authors found that their model perfectly fills a theoretical gap in Non-Continuous OFDM (where some subcarriers are null), where previous bounds were often too loose to be useful.
  • Power Distribution Sensitivity: PAPR is most severe when power is distributed evenly across the system bandwidth. If power is concentrated in a subband with fewer subcarriers, the overall PAPR of the composite signal tends to decrease.
  • Robustness to Filtering: The model remains a good approximation even when 5G-specific pulse-shaping techniques like Filtered-OFDM (F-OFDM) or Windowed-OFDM (W-OFDM) are applied, despite these techniques slightly altering the signal envelope.

Key Results

  • Superior Accuracy: In simulations with multiple numerologies (e.g., $N_1=200, N_2=100$), the proposed analytical expression (Eq. 31) showed a near-perfect match with simulation data, while older models showed significant "discernible deviation."
  • Asynchronous Support: The model successfully predicted PAPR for asynchronous systems where subbands aren't perfectly aligned in time—a common real-world 5G scenario.
  • Closed-Form Solutions: For the two-subband case, the authors provided a closed-form solution for optimal power allocation, proving that for equal-bandwidth subbands, a 50/50 power split yields the maximum mean PAPR.

Limitations and Future Work

The primary limitation is the reliance on the Gaussian approximation, which assumes a relatively large number of subcarriers. While this holds true for most 5G deployments, the model's accuracy might dip for extremely narrowband systems with very few subcarriers.

The authors suggest two main paths for future research:

  1. Developing specific PAPR reduction techniques (like clipping or selective mapping) tailored for the unique structure of mixed-numerology signals.
  2. Investigating power allocation strategies that maximize system capacity while staying under a strict PAPR constraint to protect hardware.

Summary updated

Abstract

In this paper, the probability distribution of the peak to average power ratio (PAPR) is analyzed for the mixed numerologies transmission based on orthogonal frequency division multiplexing (OFDM). State of the art theoretical analysis implicitly assumes continuous and symmetric frequency spectrum of OFDM signals. Thus, it is difficult to be applied to the mixed-numerology system due to its complication. By comprehensively considering system parameters, including numerology, bandwidth and power level of each subband, we propose a generic analytical distribution function of PAPR for continuous-time signals based on level-crossing theory. The proposed approach can be applied to both conventional single numerology and mixed-numerology systems. In addition, it also ensures the validity for the noncontinuous-OFDM (NC-OFDM). Given the derived distribution expression, we further investigate the effect of power allocation between different numerologies on PAPR. Simulations are presented and show the good match of the proposed theoretical results.
BibTeX
@article{liu2020peak,
  author = {Xiaoran Liu and Lei Zhang and Jun Xiong and Xiaoying Zhang and Li Zhou and Jibo Wei},
  title = {{Peak-to-Average Power Ratio Analysis for OFDM-Based Mixed-Numerology Transmissions}},
  journal = {IEEE Trans. Vehicular Technology},
  year = {2020},
  doi = {10.1109/tvt.2019.2960801},
}

Loading related content...

Figure